Five studies met the inclusion criteria (n=at least 1,083). Four studies recruited symptomatic and asymptomatic people and one recruited only people who were symptomatic. All studies used an appropriate reference standard, avoided verification, incorporation and clinical review biases. Four studies reported blinding of the interpreters of the index test. Four studies reported recruiting an appropriate patient population and appropriate index test methodology. Two studies avoided progression bias, reported uninterpretable results, explained withdrawals, withheld treatment until the assessments were complete and reported observer variation. None of the studies reported blinding of the interpreters of the reference standard.
Sensitivity ranged from 75% to 89%. Specificity ranged from 79% to 93%. Positive likelihood ratio ranged from 4.1 to 11.1 and negative likelihood ratio from 0.13 to 0.29. Where calculable, post-test probability of a positive test ranged from 21% to 59.7% and post-test probability of a negative test from 0.64% to 5.7%.